The convergence of generative AI and the Internet of Things (IoT) is reshaping industries, enabling real-time decision-making, predictive maintenance, and prescriptive insights. Yet, scaling these technologies across enterprises presents unique challenges from managing vast datasets to ensuring AI reliability at the edge.

As a Principal IoT Architect at SDG Group (an ALTEN company) and a dual Microsoft MVP on IoT and RTI, Sander van de Velde shares his expertise on how these innovations are transforming industrial operations.

“The most significant change we see today is the integration of generative AI with IoT, turning real-time data into predictive and prescriptive insights but doing so at scale remains the real challenge.”

What is the most significant change in IoT today, and where does the industry still struggle?

The most transformative shift in IoT is the integration of generative AI into operational workflows. We’ve moved beyond basic remote monitoring to predictive and prescriptive maintenance, where AI analyzes real-time data to forecast equipment failures or optimize processes. However, the industry still grapples with scaling these solutions.

Working with massive datasets introduces risks like AI hallucinations where models generate inaccurate predictions. Additionally, deploying generative AI at the edge on local networks with limited compute power adds complexity. We need robust frameworks to ensure reliability, especially in critical environments like offshore vessels or manufacturing plants.

Which technologies are making the biggest impact on your projects?

Two platforms stand out: Databricks and Microsoft Fabric. These tools enable seamless real-time data ingestion, replacing traditional batch processing. Clients can now transition from static reports to dynamic, AI-driven insights.

For example, we’re using operations agents as AI-powered virtual assistants, to monitor telemetry and execute predefined actions, like alerting engineers or adjusting parameters. These agents act as “virtual junior engineers”, guided by playbooks to ensure consistency. This reduces manual intervention while maintaining operational control.

Can you share a recent project where these technologies solved a critical challenge?

We recently developed a digital twin for an offshore client who needed real-time visibility into vessel operations. Previously, they relied on delayed emails and manual reports. Our solution aggregated live ship positioning, equipment telemetry, and environmental data into a unified model.

The challenge was designing a flexible rule engine to adapt to unpredictable changes like weather disruptions or equipment unavailability. By decoupling rules and leveraging edge AI, we created a system that updates dynamically. The client now gains real-time insights into action durations, improving project timelines and resource allocation.

What’s the biggest hurdle organizations face when scaling generative AI in IoT?

Scaling generative AI from proof-of-concept to enterprise-level deployment is the primary obstacle. Small-scale pilots succeed with curated datasets, but real-world applications involve raw, unstructured data at scale. This introduces risks like hallucinations, security vulnerabilities, and performance bottlenecks.

To mitigate these, we focus on:

• Better playbooks and guardrails to guide AI agents.
• Hybrid data strategies, combining real-time telemetry with contextual datasets (e.g., ontologies for equipment hierarchies).
• Edge optimization, ensuring AI models run efficiently on local hardware.

How has ALTEN’s approach to IoT evolved in the past five years?

Initially, our focus was on moving data from devices to the cloud, a technical challenge requiring expertise in industrial protocols and cloud integration. Today, we’ve shifted to turning raw telemetry into real-time insights using architectures like the medallion model.

Now, we’re pushing further into predictive and prescriptive maintenance, helping clients anticipate failures before they occur. This evolution reflects ALTEN’s commitment to bridging OT (Operational Technology) and IT seamlessly, enabling industries to operate smarter, faster, and more resiliently.

Description: Discover how generative AI and IoT are revolutionizing industrial operations, from predictive maintenance to real-time insights.